WIMP Dark Matter

Why anyone proposed a new particle in the first place

By the late 1970s the observational case for dark matter had become essentially decisive. Vera Rubin and Kent Ford's photometric and spectroscopic rotation curves showed that the orbital speed of gas and stars in spiral galaxies did not fall off as 1/√r the way Keplerian dynamics around the visible mass would predict — instead it flattened, requiring roughly five times more enclosed mass than the luminous component beyond the optical disk. Cluster-scale virial mass estimates from Fritz Zwicky going back to 1933 demanded the same. Gravitational lensing surveys, the CMB acoustic-peak structure, and the matter-power spectrum at every observable scale have since converged on Ω_DM ≈ 0.27 of the cosmic energy budget — about 84 % of the matter is non-baryonic, cold, and so far seen only gravitationally.

The Standard Model of particle physics does not contain a viable dark-matter candidate. Neutrinos are too light (m_ν ≲ 1 eV, hot and fast) to clump on galaxy scales. No stable Standard-Model particle has the right relic abundance. Some new particle is required, and the question is what mass and coupling it has.

The WIMP miracle

The leading hypothesis from the early 1980s onward was that the dark-matter particle is a stable thermal relic of the early universe — a heavy particle in equilibrium with the Standard-Model plasma at very high temperatures, dropping out of equilibrium when its annihilation rate falls below the Hubble expansion rate. The frozen-out comoving abundance depends almost entirely on the annihilation cross-section ⟨σv⟩ at decoupling:

Ω_DM h² ≈ 0.1 × (3 × 10⁻²⁶ cm³/s) / ⟨σv⟩

If you set ⟨σv⟩ to the value naturally produced by a particle with electroweak-scale mass M ≈ 100 GeV and weak-scale coupling — schematically σ ~ G_F² M² with G_F the Fermi constant — you get

⟨σv⟩_weak ≈ G_F² M² · v ≈ 10⁻²⁶ cm³/s   (M ≈ 100 GeV)

and an abundance Ω_DM h² ≈ 0.1 — astonishingly close to the measured value of 0.120 ± 0.001. This coincidence between (a) the dark-matter density actually observed and (b) the relic density predicted from particle physics aimed at the electroweak hierarchy problem (supersymmetry, extra dimensions, little-Higgs models) is the "WIMP miracle." It was the single most compelling theoretical argument for dark-matter physics for thirty years.

Thermal freeze-out — the actual calculation

The Boltzmann equation governing a relic abundance n(t) of a stable species χ in an expanding plasma is

dn/dt + 3 H n = -⟨σv⟩ (n² - n_eq²)

While the plasma is hot (T ≫ m_χ), n ≈ n_eq is fixed by thermal equilibrium. As temperature drops below m_χ, n_eq falls exponentially like e^(-m_χ/T) (Boltzmann suppression). Annihilation rate Γ = n ⟨σv⟩ tracks this falling density. Freeze-out occurs when Γ ≈ H — the universe expands faster than the χ can annihilate — and the comoving number density "freezes" at its instantaneous value. For typical WIMPs this happens at T_f ≈ m_χ/25. The frozen-out abundance scales like 1/⟨σv⟩, hence the inverse relation above. Cosmologists call this picture the "thermal-relic abundance" or "freeze-out scenario," and it is a 50-year-old fixed point of the standard cosmological picture.

Concrete particle candidates

The WIMP miracle is generic — it doesn't specify what the particle is. Several well-motivated theories of physics beyond the Standard Model contain a stable, heavy, weakly-interacting particle as a free outcome. Three families dominate.

• Neutralino LSP. R-parity-conserving supersymmetry contains a "lightest supersymmetric particle" that is stable and a Majorana fermion. The most common LSP is the neutralino χ⁰₁ — a quantum-mechanical mixture of the superpartners of the B (bino), W³ (wino) and the two Higgs doublets (higgsinos). Different mass-eigenstate compositions give very different phenomenology. Pure bino has small annihilation rate and typically over-closes the universe unless something boosts it (coannihilation, resonance). Pure wino at m ≈ 2.9 TeV and pure higgsino at m ≈ 1.1 TeV both give the right relic density via thermal freeze-out and remain viable above LHC reach.
• Kaluza-Klein dark matter. In universal extra-dimensional models with the inverse compactification scale R⁻¹ at a TeV, the lightest Kaluza-Klein excitation of a Standard-Model field is stabilised by Kaluza-Klein parity. The first KK excitation of the photon (or hypercharge gauge boson, B¹) at m ≈ 800-1300 GeV is a viable WIMP, with annihilation, recoil and indirect signatures qualitatively similar to but quantitatively distinct from supersymmetric ones.
• Singlet-scalar / Higgs-portal WIMP. A minimal extension of the Standard Model with a single real scalar S coupled to the Higgs as λ|H|² S² gives a stable WIMP if λ_HS is small and the model has a Z₂ symmetry. Its relic abundance, direct-detection cross-section, and Higgs-invisible-width are all set by a single coupling — easy to constrain and probe in tandem.

Three detection channels

The same coupling that gives the WIMP miracle also lets us look for WIMPs in three complementary ways, all driven by the schematic interaction χ χ ↔ q q (or χ χ ↔ ℓ ℓ, χ χ ↔ W W, etc):

Channel	Process	Reads	Flagship experiments
Direct	χ + N → χ + N (elastic)	Galactic-halo WIMPs scatter off a nucleus in a deep-underground detector. Recoil energy ~ keV.	XENONnT, LZ, PandaX-4T, SuperCDMS, DAMIC, DEAP-3600
Indirect	χ + χ → SM SM	WIMP pair annihilation in dense regions (Galactic Centre, dwarf spheroidals, the Sun) produces γ-rays, ν, e⁺, p̄.	Fermi-LAT, HESS, MAGIC, VERITAS, CTA, IceCube, AMS-02
Collider	p p → χ χ + X	Direct production at the LHC, recoiling against an ISR jet, photon or Z. Missing transverse energy ≡ WIMPs.	ATLAS, CMS (monojet, monophoton, mono-Z)

The triumph of WIMP phenomenology in the 2000s was that all three channels probe overlapping but distinct parameter regions, so the WIMP hypothesis is testable from many directions at once. The crisis of the 2020s is that none of those channels has produced a signal.

Direct detection — the underground race

The signal a direct-detection experiment hunts is a low-energy (1-50 keV) nuclear recoil produced when a halo WIMP scatters elastically off a target nucleus. The expected event rate for a benchmark spin-independent (SI) cross-section σ_SI is roughly

R ≈ 1 event / (tonne · year) × (σ_SI / 10⁻⁴⁵ cm²)

which means tonne-scale exposures are needed to test the cosmologically interesting σ_SI ~ 10⁻⁴⁶ cm² and smaller. The state of the art is the dual-phase liquid-xenon time-projection chamber (LXe TPC): a tank of ultra-pure liquid xenon (typically 5-7 tonnes of active mass) instrumented with photomultiplier arrays above and below, and a strong electric field that drifts ionisation electrons from a recoil up into a gas phase where they produce a delayed proportional-scintillation pulse (S2). The prompt scintillation pulse (S1) and the delayed S2 give a 3-D position, an energy, and a ratio S2/S1 that discriminates nuclear recoils (signal) from electron recoils (β/γ background).

Experiment	Target	Active mass	Best σ_SI bound	Status (2026)
LUX (2013-2016)	LXe TPC	250 kg	~10⁻⁴⁶ cm² @ 50 GeV	Decommissioned
XENON1T (2016-2018)	LXe TPC	2 t	4.1 × 10⁻⁴⁷ cm² @ 30 GeV	Decommissioned
PandaX-4T	LXe TPC	4 t	~3 × 10⁻⁴⁸ cm² @ 30 GeV	Running
XENONnT	LXe TPC	5.9 t	~2 × 10⁻⁴⁸ cm² @ 30 GeV	Running
LZ (LUX-ZEPLIN)	LXe TPC	7 t	~1.5 × 10⁻⁴⁸ cm² @ 30 GeV	Running
DARWIN / XLZD	LXe TPC	40-50 t	~10⁻⁴⁹ cm² (projected)	Late 2030s
SuperCDMS	Ge / Si bolometer	kg-scale	~10⁻⁴³ cm² @ 5 GeV (low-mass)	Running
DEAP-3600	LAr single-phase	3.3 t	~10⁻⁴⁶ cm² @ 100 GeV	Running

The composite picture across two decades is dramatic: the spin-independent cross-section bound at m_χ ≈ 30 GeV has dropped by roughly three orders of magnitude since 2010 with no signal. Plotted on the canonical σ_SI vs. m_χ diagram, the excluded region has marched downward, swallowing nearly all of the "natural" supersymmetric parameter space.

Indirect detection — annihilation signatures from the halo

If WIMPs annihilate to Standard-Model particles, then anywhere they are dense enough they will leak energetic photons, neutrinos and cosmic rays. The expected differential flux is

dΦ/dE = (1/4π) · ⟨σv⟩/(2 m_χ²) · dN/dE · J

where J = ∫ ρ² dℓ is the line-of-sight integral of the WIMP density squared, dN/dE is the prompt photon spectrum per annihilation, and ⟨σv⟩ is set by the WIMP miracle to its thermal-relic value. The dimensional argument: take a 100 GeV WIMP annihilating to bb̄ in a region of mean density 0.4 GeV/cm³ — flux is testable with γ-ray telescopes targeting dense regions.

• Galactic Centre. Highest J factor on the sky but enormous astrophysical background (the diffuse γ-ray emission and ~10⁵ point sources). The "Galactic-Centre excess" — an unexplained 1-3 GeV bump seen by Fermi-LAT and interpreted by some authors as ~30 GeV WIMPs annihilating to bb̄ — has gone back and forth as candidate evidence. Most current analyses prefer an unresolved millisecond-pulsar population over WIMP origin, but the question is not closed.
• Dwarf spheroidal galaxies. 30+ companions of the Milky Way are essentially dark-matter-dominated (M/L ratios of order 100) and have minimal astrophysical γ-ray emission. Stacked Fermi-LAT analyses of dwarfs give the cleanest WIMP annihilation bounds. As of 2024-2025, they exclude the thermal-relic cross-section ⟨σv⟩ = 3 × 10⁻²⁶ cm³/s for m_χ ≲ 100 GeV in the dominant annihilation channels — a sharp, unambiguous null.
• Galaxy clusters and the cosmological background. Larger but more diluted J factors. Useful at higher m_χ where CTA (Cherenkov Telescope Array, coming online 2027-2029) will set best limits.
• Sun. WIMPs scattering in the Sun lose enough energy to be gravitationally captured, accumulate in the core, and annihilate to a νν̄ flux escaping outward. IceCube searches this — limits competitive with direct detection for spin-dependent couplings.
• Cosmic-ray positrons and antiprotons. AMS-02 on the ISS sees an unexplained excess of positrons above 10 GeV. Initially interpreted as possible WIMP annihilation, the spectrum and the simultaneous lack of an antiproton excess strongly favour pulsar origins.

Collider production — monojet + missing energy

If WIMPs exist with electroweak-scale couplings, the LHC can produce them in proton-proton collisions, recoiling against initial-state-radiation jets, photons or Z bosons. The signature is monojet + missing transverse energy (MET) — a single hard jet with no balancing visible activity. ATLAS and CMS have run extensive monojet searches, increasing MET cuts and energies through Run 1, 2 and 3. No signal. As of the latest data, the LHC excludes squarks lighter than ~2 TeV in simplified models with massless neutralinos, and gluinos below ~2.5 TeV. Direct WIMP production via effective operators or simplified models with explicit mediators (Z′, scalar mediator, t-channel sleptons) sets independent constraints; the LHC excluded region overlaps significantly with the direct-detection excluded region in benchmark models, eliminating the most natural slice of parameter space twice over.

Worked example — does a 100 GeV WIMP give the right Ω?

Take m_χ = 100 GeV and an annihilation cross-section set by the weak coupling, σ ≈ G_F² m_χ². Plugging in G_F = 1.166 × 10⁻⁵ GeV⁻²:

σ ≈ G_F² m_χ² = (1.166e-5)² × (100)²  GeV⁻²
              = 1.36e-6  GeV⁻²
              = 1.36e-6 × (0.197 fm)²  (using 1 GeV⁻¹ ≈ 0.197 fm)
              ≈ 5.3e-8  fm²
              ≈ 5.3e-34  cm²

⟨σv⟩ ≈ σ × v ≈ 5.3e-34 cm² × (c/3) ≈ 5e-24 cm³/s
                     (decoupling velocity ~ c/3 → v ≈ 1e10 cm/s)

This is two orders of magnitude too large — but we have made the simplest dimensional estimate. A more careful estimate at the decoupling temperature T_f ≈ m/25, accounting for thermal averaging, partial-wave structure, and the running of the weak coupling, gives ⟨σv⟩ closer to 10⁻²⁶ cm³/s — within a factor of a few of the measured thermal-relic value. That is the WIMP miracle in arithmetic form: an order-of-magnitude estimate using only the Fermi constant and the electroweak scale reproduces the cosmologically measured dark-matter abundance to within a factor of three. No fine-tuning, no new parameters. This is why the WIMP was the favourite for forty years.

The "WIMP crisis"

By 2014 the canonical electroweak WIMP parameter space had been substantially squeezed. By 2020 most of the simplest models were excluded. By 2025 the situation is what working dark-matter physicists openly call the "WIMP crisis":

• The direct-detection σ_SI bound at m_χ ≈ 30 GeV is below 1.5 × 10⁻⁴⁸ cm² (LZ 2024), three orders of magnitude below the canonical pre-LHC WIMP prediction.
• The Fermi-LAT dwarf-spheroidal stacked analysis excludes ⟨σv⟩ at the thermal-relic level for m_χ ≲ 100 GeV in the bb̄ and τ⁺τ⁻ channels.
• The LHC, after 4000 fb⁻¹ of total integrated luminosity, has found no superpartners up to m ~ 2 TeV. "Natural" supersymmetry — the variant motivated by the electroweak hierarchy problem and most likely to produce a WIMP — is increasingly fine-tuned.
• Direct-detection sensitivity will reach the irreducible neutrino floor at σ_SI ≈ 10⁻⁴⁹ cm² in the early 2030s with DARWIN/XLZD-class 40-50 tonne detectors. Below the floor, each factor of two in σ costs orders of magnitude in exposure — the canonical search effectively ends.

A "crisis" in this technical sense doesn't mean the WIMP is excluded — only that the most natural slice of WIMP parameter space, the slice that originally motivated the four-decade investment, is gone. Heavy-WIMP targets above LHC reach (pure-higgsino at 1.1 TeV, pure-wino at 2.9 TeV) remain viable but require a 100 TeV pp collider (FCC-hh) or a high-energy muon collider to probe directly. The community is rotating attention.

Alternatives gaining ground

Candidate	Mass	Motivation	Leading experiments
QCD axion	1 μeV – 1 meV	Peccei-Quinn solution to strong-CP	ADMX, HAYSTAC, ABRACADABRA, DMRadio
Sterile neutrino	~ keV	Neutrino oscillations, 3.5 keV X-ray line	X-ray telescopes (XRISM, NewAthena)
Primordial black hole	10¹⁷ g window	Inflation-era density perturbations	Microlensing, GW (LIGO/LISA)
Fuzzy / ultralight scalar	~ 10⁻²² eV	Solves small-scale-structure puzzles	Pulsar timing, Lyα forest, galactic dynamics
Dark-sector / hidden photon	MeV – GeV	Self-interacting DM, anomalies (g-2)	FASER, SHiP, fixed-target experiments
Heavy WIMP (higgsino/wino)	1 – 3 TeV	Still naturally relic-abundance correct	FCC-hh (proposed), CTA, future muon collider

None of these is yet a "discovery" either. But the QCD axion in particular benefits from the same compelling theoretical motivation the WIMP once enjoyed — it solves a different but equally important problem (strong-CP) and naturally has dark-matter-relevant abundance — and unlike the WIMP, the entire well-motivated axion-window mass range is technologically accessible to the next decade of haloscopes.

Where WIMPs show up in observational astronomy

Even without a particle-physics detection, dark matter — whatever it is — is everywhere in modern astrophysics. WIMPs are the default assumption in:

• N-body simulations. Millennium, IllustrisTNG, FIRE-2, EAGLE all assume cold, collisionless dark matter — phenomenologically equivalent to a heavy WIMP, but the same simulations work for any candidate with σ_self/m ≪ 1 cm²/g and v_dispersion small. Self-interacting dark matter (SIDM, σ/m ~ 1 cm²/g) is an alternative now being seriously simulated.
• Galaxy rotation curves. A Navarro-Frenk-White (NFW) profile fit to a flat rotation curve, with parameters set by cosmological simulation, is the default WIMP halo model. The "core-cusp" tension at the centres of dwarfs is the most-discussed challenge.
• Gravitational lensing. Weak-lensing surveys (DES, KiDS, HSC) map projected dark-matter density on the sky. The substructure visible in strong-lens systems probes the dark-matter clustering down to ~10⁹ M☉ — favouring cold-WIMP-like over warm dark matter at present.
• CMB acoustic peaks. The peak ratios and matter-power-spectrum constraints from Planck (Ω_DM h² = 0.120 ± 0.001) are agnostic between cold WIMPs and other cold candidates but exclude hot or strongly self-interacting candidates.
• Bullet Cluster and analogues. Direct gravitational-lensing mapping of cluster collisions shows dark-matter offset from collisionless hot gas — confirming the dark-matter component is essentially collisionless on cluster crossing timescales (σ_self/m ≲ 1 cm²/g), consistent with WIMPs.

Future of the WIMP — endpoints in sight

Two experiments will set the endpoint of the canonical WIMP search:

• DARWIN / XLZD. A merged 40-60 tonne LXe TPC concept (consolidation of the XENON, LZ and DARWIN collaborations, called XLZD or "DARWIN") targeting σ_SI ~ 10⁻⁴⁹ cm² in the early 2030s. This sensitivity touches the neutrino floor — diffuse-supernova, atmospheric and ⁸B-solar neutrinos coherently scatter on Xe at recoil energies indistinguishable from WIMPs. Beyond the floor each factor of two in σ costs orders of magnitude in exposure.
• CTA (Cherenkov Telescope Array). Coming online 2027-2029. Order-of-magnitude better sensitivity to TeV γ-rays than HESS or MAGIC. Will set the strongest indirect-detection bounds on heavy-WIMP annihilation in the Galactic Centre, dwarf spheroidals and clusters, probing the m_χ = 1-10 TeV "heavy WIMP" target region currently untouchable by direct detection.

If neither sees a signal by ~2035, the canonical WIMP — the thermal-relic with mass 10 GeV-10 TeV and weak-strength couplings — is excluded across essentially all of its naturally motivated parameter space. The remaining viable WIMP corners (multi-TeV pure higgsinos and winos) await a 100 TeV proton-proton collider or a high-energy muon collider, neither of which will deliver data before the 2040s.

Common pitfalls

• Confusing "dark matter" with "WIMP". Dark matter is the observed gravitational phenomenon. WIMP is one (now strained) hypothesis for what particle does the dark-matter job. Ruling out WIMPs does not rule out dark matter.
• Confusing σ_SI and σ_SD. Spin-independent cross-sections benefit from coherent A² enhancement on heavy targets like xenon; spin-dependent bounds are weaker but probe different physics. Bound papers usually plot both.
• Treating the GeV Galactic-Centre excess as confirmed WIMP signal. Most current analyses prefer unresolved millisecond pulsars; the excess remains a hint, not a detection.
• Forgetting the neutrino floor is not a hard wall. It is a transition from background-free to background-dominated counting — extracting a WIMP signal is still possible above the floor with directional or annual-modulation handles, just much harder.
• Identifying "WIMP" with "neutralino". The neutralino is the most-studied WIMP candidate but not the only one. Kaluza-Klein, singlet-scalar, dark-photon-mediated, and many other models give WIMP-class phenomenology with different details.
• Assuming non-detection rules out supersymmetry. Even if the neutralino is excluded as a viable WIMP, SUSY breaking and the broader SUSY parameter space contain many other LSP candidates (gravitino, axino) that are not WIMPs in the technical sense. WIMP exclusion is a strong constraint on natural SUSY, not on the whole framework.